Numerical Study Using NARX to Improve Pipe Vibration in Dosing Pump Systems

Authors

Shmoos Rahem Mazal, Department of Mechanical Engineering, College of Engineering, Al-Nahrain University, Jadriya, Baghdad, Iraq AND Collage of Electromechanical Engineering, Department of Electromechanical Systems Engineering, University of Technology, Baghdad, IraqFollow
Mohammed Jawad Mohammed, Collage of Electromechanical Engineering, Department of Electromechanical Systems Engineering, University of Technology, Baghdad, Iraq
Mahmoud Mustafa Mahdi, Collage of Electromechanical Engineering, Department of Electromechanical Systems Engineering, University of Technology, Baghdad, Iraq

Abstract

Pressure pulsation is a serious problem in the piping system. It is generated at the beginning and end of the water-pumping process, or when valves suddenly close. The high turbulence of fluid particles in the presence of pressure waves causes multiple stresses on the pipe walls, requiring maintenance and eventual replacement over time. In this research, an experimental setup for a piping system vibration model induced by pulsating pressure is designed, developed, and fabricated. The objective of the present work is to create a MATLAB program using the NARX method at the beginning of the Neural Network time series. Nonlinear Auto-Regressive External Input was described as a method of system identification to predict pressure pulsation and acceleration as a performance measure for the system. NARX can be used to model and predict flow rates based on the input (flow rate) and output (acceleration, pressure) in four cases. For the fourth case (3/4 closed valve), the pressure pulsation initially reaches 0.8 bar and is reduced to 0.3 bar after applying the NARX model in Simulink, corresponding to a 75% reduction. Similarly, the fourth case (3/4 closed valve) yields the best acceleration response. The pressure is 0.1 m/S2 and is reduced to 0.007 m/S2 after using NARX in Simulink. The percentage improvement in acceleration after using NARX in Simulink is 93%.

Recommended Citation

Mazal, Shmoos Rahem; Mohammed, Mohammed Jawad; and Mahdi, Mahmoud Mustafa (2026) "Numerical Study Using NARX to Improve Pipe Vibration in Dosing Pump Systems," Engineering and Technology Journal: Vol. 44: Iss. 1, Article 15.